NVIDIA’s PCI-Express Graphics Family

Teehee.

 

Not too big I hope.

 

Acutally - you did say:
No one said it isn't legitimate. It's just not as optimal as ATI's.

If that is not a definitive judgement, I don't know what is...

About your url.

My apologies. And I do mean that. IT was a figure of speech. In other words, provide the information regarding the bridge performance vs native support.

I was not asking to see your house, car, airplane, whatever. I honestly do not care. You can live in a tin can for all I care - doesn't make me think more or less of you.

 

What if i live in a bottle?

 

Does anyone really dispute that? That's just more of a common sense thing. I mean, is a native PCI-E implementation going to be worse than AGP?

The only question is about "realized" performance improvement. And no one here has made a definitive judgement about that.

Apology accepted...my 'betting my house' was also a figure of speech.

Again, it's simply a difference between theoretical and "realized" performance gains. You seem to be claiming that we're all concluding that ATI's native solution to show significant realized performance increases. That's not what we're saying. I'm not "beting my house" that the native solution gives a REALIZED performance increase. I would bet that would show no realized difference, at worst.

Nor should it.

 

Off Topic Warning:

Aaah, Canon in D, i just played that song on my guitar. Well, more like tried to play it

 

Give me a break, I know exactly what he's saying it just happens to be a piss poor point.
You nor he can make any kind of point on the subject until both solutions are benchmarked, all we CAN say is that bridge will have NONE of the advantages of native PCI-Express. Now take a leap into a deep dark hole until both are benchmarked guest it'll keep your brand of fan-boi rhetoric silent until that point, which I'm sure will be music to all sane peoples ears.

 

Total hypocrisy coming out of your mouth duncan. If you do "find all posts by duncan36", a quick analysis will show that you have been making prognostications about NV40 performance for a while now instead of "waiting" to see the benchmarks. But more than that, your posts show a distinct, aggressive, fanboi bias against NVidia.

So don't go telling other people to wait for evidence when you are not willing to do so yourself.

 

I bought a Sony TRV900 in 2000 to film our wedding. The reason I asked is because I was wondering how much you were likely to spend on video quality and a film hobby. If you told me you bought a PD150, VX2000, or XL1, I'd be apt to believe you had a real need for an HDCAM. Right now, HD cams are nothing more than expensive toys and will remain for a few years. Today's HDCAMs have expensive CCDs paired up with crappy lenses, and they cost $4000+. To really do serious HD editing, you'd need to plunk down probably $10,000 for the camera, and a few thousand more for an NLE system (enough to hold all the video). I just don't see it as realistic.

Yes, I sounds like a cool thing to do: edit HD video. The reality is, producing SD video is already hellishy expensive and difficult, and if your not serious about doing it, you HD CAM video will look like crappy Mexican soaps. Sure, the resolution will be higher, but it will have artifacts, blurriness, and really shitty lighting. It's like people who buy 10 megapixel point-and-shoot cameras, and produce stunningly high resolution utterly crappy shots that look no better than 1 megapixel images.

I believe you have a desire to produce HD high quality video for a big screen TV in your home, I just question whether or not you are ready to commit the financial and time resources needed to utilize the tools and make a difference. It would be a real shame to spend $10k on a nice camera, and produce videos with no better quality than a cheap $500 camcorder.

 

ExtremeTech article up:

nVidia Talks Up PCI Express at IDF

MuFu.

 

I don't have a "real need" for the $2500 or so I justr spent on my computer. My Tualatin 1.4 was perfectly "adequate."

I had certain requirements for my last cam: it had to be relatively compact, (no disincentive not to take it places) and it had to use removable flash media for taking stills. (Was not a "common" feature 4 years ago, and we did not want to buy a separate still digital camera then). Of course, optics and all that played a part as well.

Right at this point in time...probably not. That's sort of the whole purpose of the ATI and Pinnacle announcement though...it's going to get cheaper (at least the editing part)...and soon.

As for cameras, we'll have to wait and see.

 

Thanks for putting me in my place. I'll crawl back in my deep dark hole as you told me toll...Anyway /ignore

I am a fan of neither, and I own several of both - what does that have to do with my opinion.

Actually, the discussion with Mr. DeFuria ended ok - or so I thought.

 

Please...call me Joe.

 

Joe, I can respect a man who buys expensive HW just because he wants to have the best. I'm the same way. I dabble in amateur photography, but I spent far too much on lenses for my EOS-10D, and I am in no position to noice the different between a consumer lens and a professional lens. I just don't like buying gadgets that I know are not the best price/performance.

I just wanted to point out that HD editing is super razor bleeding edge right now. It's like ATI trumping the fact that you could network 256 R300's together and do offline rendering with them. Yeah, it's a nice idea, and pretty cool, but guess what? No one is building renderfarms out of 3d cards yet. Vast majority of 3d CGI will be produced on standard linux renderfarms for forseeable future.

There are plenty of reasons to buy an R420, and plenty of reasons to like PCI-Express, but criticizing anyone who had a poor implementation at this point in time is rather premature.

Anyway, AGP8x was supposed to have isochronous transfers and faster-than-pci framebuffer reads, why don't any AGP implementations seem to support this? "Native" PCIE implementations may suffer similar problems, with first implementations being buggy, or lacking full support of all features.

 

That's all well and good...I just don't like the implication that I did not mean what I said.

 

Why does it matter that HD video editing is cutting edge? Obviously that's how all technology starts out. Once it hits mainstream consumer prices and gets some stable, user-friendly software backing it up, it becomes mainstream. Remember, we are talking about bus technologies and GPUs that aren't even on the market yet! Why is it so far-fetched to think HD video editing won't take off within the lifecycle of the upcoming PCI Express products? A lot of very big companies are putting their marketing muscle behind that very idea.

And by the way, did you ever stop to think that the lack of a high speed bidirectional interconnect like PCI Express may be one of the things holding back GPU render farms from being feasible? Check into some of the cluster processing research that's been appearing at recent SIGGRAPHs... it's virtually begging for something like this.

As soon as information on an upcoming product becomes available, then it's subject to praise or criticism. Nvidia has presented information which raises some serious questions about their PCI Express support. According to Extremetech, their "demonstration" of this technology at IDF consisted of a simple HD video playback, which practically any system today can handle, and certainly does not require PCI Express. At least ATI demonstrated something that ran significantly better on their PCI Express hardware.

Well, like I just said, ATI has already shown that they can get a practical application to take advantage of the higher performance of PCI Express. I don't think the same could be said of anyone (including ATI) when comparing AGP 8X to AGP 4X during its entire lifecycle.

 

Well meet, Joe. My name is Scott. /shake

 

It's not even cutting edge, it's vapor for consumers. I'm not arguing about the desirably of HD editing, I'm talking about whether it will be a reality for any market segment before NVidia ships an NV40 with native PCI-E (assuming their bridge even lacks the capability to provide HD readback)